bonded repair: design and process considerations · – reliance on bonds for flight safety...
TRANSCRIPT
1
Bonded Repair:Design and Process Considerations
Michael Borgman ~ Spirit AeroSystemsCo-Chair ~ Commercial Aircraft Composite Repair Committee
CACRC Spring 2007 ~ May 7-11 ~ Amsterdam
2
Introduction
• Generally, bonded repairs are not allowed in structurally critical applications
– Reliance on bonds for flight safety generally disallowed• One reason is strength performance scatter resulting from the somewhat limited precision inherent to bonded repair design, analysis, and processing
• A second is the absence of a means of verifying bond strength in the finished product
• This presentation provides thoughts on ways to enhance precision [of process control]
3
Introduction• Any scenario for process control must be evaluated for reliability of
success (probability must be assessed)• This reliability must be demonstrated through mechanical testing the
product from a LARGE volume of bondment process cycles• Have started this process at SPIRIT in support of one of our products,
and have tested scarf joint strengths from approximately 200 bondment process cycles
– Many different mechanics– Many different cure thermal profiles– Fresh and Old Material– Many different resulting NDI scan quality results
• Have learned there are nuances to be considered in selecting a joint geometry for process reliability testing
– Joint shape and edge proximity (breathing during cure)
4
Introduction
• This presentation is focused on bonded joints– Discussion topics
1. Regulations / Requirements2. Design and analysis
a) Joint type comparisonsb) Parametric trends
3. Surface preparationa) Geometry creationb) Pre-bond surface cleanlinessc) Pre-bond substrate moisture content
4. Laminate patch preparationa) Precision ply cutting
5. Post-bond cure state assessment
5
FAR Regulations• Design Requirements• 14 CFR 23.305 / 25.305 Strength and Deformation
– Support limit loads without detrimental permanent deformation– Support ultimate loads without failure for at least 3 seconds
• 14 CFR 25.571 Damage Tolerance and Fatigue Evaluation of Structure– Evaluation of strength, detail design, and fabrication must show
• Catastrophic failure due to fatigue, corrosion, manufacturing defects, or accidental damage, will be avoided
• Capability to successfully complete a flight during which likely structural damage occurs– [Impact, lightning (25.581), etc…]– Damaged structure able to withstand the static loads (considered as ultimate loads) which are reasonably
expected– Each evaluation must…
• [Include] typical loading spectra, temperatures, and humidities• [Be based on] analysis, supported by test evidence• [Make the assumption that] structure contains an initial flaw of the maximum probable size that
could exist as a result of manufacturing or service-induced damage
6
FAR Regulations• Design Requirements (cont’d)• 14 CFR 23.573 Damage tolerance and fatigue evaluation of structure(a) Composite Airframe Structure
– For any bonded joint, the failure of which would result in catastrophic loss of the airplane, the limit load capacity must be substantiated by one of the following
1. Maximum disbonds of each bonded joint consistent with the capability to withstand the loads in paragraph (a)(3) of this section must be determined by analysis, tests, or both. Disbonds of each bonded joint greater than this must be prevented by design features
2. Proof testing must be conducted on each production article 3. Repeatable and reliable non-destructive inspection techniques must be
established that ensure the strength of each joint
7
FAR Regulations• Design Requirements (cont’d)14 CFR 23.573 Damage tolerance and fatigue evaluation of structure(a) Composite Airframe Structure
– Demonstrate by tests, or by analysis supported by tests, structure capable of carrying ultimate load with damage up to the threshold of detectabilityconsidering the inspection procedures employed
– Growth rate or no-growth of damage from fatigue, corrosion, manufacturing flaws or impact damage, under repeated loads expected in service, must be established by tests or analysis supported by tests
– Requirements on residual strength with damage1. Critical limit flight loads with the combined effects of normal operating pressure and
expected external aerodynamic pressures 2. The expected external aerodynamic pressures in 1g flight combined with a cabin
differential pressure equal to 1.1 times the normal operating differential pressure without any other load
8
•“Scarf Lap Joint” (joint without load eccentricity)
•“Scarf Rate” = L / t
Joint GeometriesBasic Definitions: “Scarf Joint”
RepairSubstrate
Adhesive
L
tRepairSubstrate
Adhesive
L
t
Substrate
Substrate
Repair
Repair
9
•“Step Lap Joint” (joint w/moderate load eccentricity)
•“Step Rate” = L / t = “Per Ply Overlap” / “Ply Thickness”
Joint Geometries Basic Definitions: “Step Joint”
L
t
Substrate
Substrate
Patch
Patch
SubstratePatch
10
•Poor design details = “square cut” edges on damage removal and/or patch– Large bond stress riser at square edge
Joint Geometries Basic Definitions: Poor Joint Details
AA
AA
A-A
Substrate
PatchA-A
Substrate
Patch
11
Design Preface“Achilles Heel” Of Bonding
• Cannot verify bond strength via post-cure inspection• In practice, rigorous control of materials and processes is
approach used to ensure bond strength in final product• Post-cure NDI can identify some quality problems
– Porosity, delamination, inclusion, non-bond• However, “kissing bond” may not be found, and…• Fundamentally, the bond strength is unknown• As a result…
The structural integrity of any bond can be questioned
12
Design Preface“Achilles Heel” Of Bonding
• Absence of “NDI bond strength guarantee” limits scope of bonded repair applications
• These limits ARE necessary for today’s level of technology (if can’t prove it’s good, then must assume it’s not)
∴In response, the repair design philosophy is…Continued safe flight with repair completely failed
– Ultimate load capability restored by presence of repair– However, safe flight (limit) is not dependent on repair
13
Analysis PrefaceHistoric Perspective
• Analysis not generally performed on bonded repair joints• Instead, substantiation data is developed for specific joint geometry
(point design data)• The substantiated joint is then used without analysis• Substantiated joint generally chosen such that ultimate failure occurs
in adherends not in adhesive bond-line• Damage tolerance analysis not generally performed• “Point design” repair data approach has worked well for existing
relatively simple composites architectures• Next generation composite airframe structures are more complicated
architecture with more complex loading• Broader array of point design data will be required to cover the
breadth of potential application needs• May be impractical
14
Analysis Preface (continued)Approach Shift For Today’s Materials
• Today’s very strong adherend laminates sometimes require joint geometries with large joint overlaps (to fail adherends instead of adhesive bond)
• These large overlaps can sometimes significantly complicate repair geometry by impinging on surrounding structural details
• “Good” bonded joints display relatively repeatable failure stresses (peel/shear)
• From structural integrity perspective need predictable, repeatable, failure mode plus understanding of joint static strength, durability, and damage tolerance
• If sufficient data is created and if analysis method can be developed and proven effective ~ ”Is design for bond failure” a reasonable option to consider?
15
• General steps in bonded scarf repair (flush repair shown)1. Design and analyze repair2. Damage removal3. Prepare substrate repair geometry
• Scarf joint shown but other joint types exist4. Create patch (ply map)5. Clean surface6. Apply adhesive7. Lay-up patch8. Cure and inspect
Repair Process Steps
-45
-4545
900
090
45-45
-4545
900
090
45
-45
-4545
900
090
45-45
-4545
900
090
45
16
Repair “Joint” Analysis
17
Repair “Joint” AnalysisTaper Region Stiffness Decay
• Laminate stiffness decays in non-linear fashion in joint step or scarf (taper) region
• Linear stiffness variation only present if all plies same stiffness (i.e., uni-directional laminate)
• Analysis methods which neglect laminate stacking sequence cannot predict stress/strain distribution
18
Analysis Preface (continued)Taper Region Stiffness Decay
• Consider [90/45/0/-45/90/45/0/-45]s “all tape” laminate (25/50/25)
– Non-linear stiffness decay as “Step” taper is traversed10
0%
99%
95%
80%
75%
74%
70%
55%
50%
45%
30%
26%
25%
20%
5% 0.8%
0%20%40%60%80%
100%120%
16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1
Lam
inat
e St
iffne
ss
(lb/in
)
Ply Count
19
• Tapered “Scarf” joint displays similar behavior – Non-linear stiffness decay as “Scarf” taper is traversed
Analysis Preface (continued)Taper Region Stiffness Decay
20
Analysis Preface (continued) Adhesive Elastic-Plastic Behavior
• Epoxy Displays Elastic-Plastic Deformation– However, Behavior May Not Be Reliably Repeatable– Treating As Purely Elastic Sometimes Appropriate
Courtesy: NIAR
21
Analysis and DesignJoint Selection Considerations
• Adherend strain profile divided by far-field strain– Common adherends in all [90/45/0/-45/90/45/0/-45]s
• Step lap yields highest far-field strain multiplier– Adherend max strain ≈ 250% far-field strain
» Applies to specific configuration analyzed only
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
0
1
2
3
P
P
P
P
Tension
22
• Bond shear stress profile at equal load (normalized)– Same adherends in each case shown: laminate = [90/45/0/-45/90/45/0/-45]s
– Neglecting plasticity effects, stress 50% higher in step, 350% higher in square
0
1
2
3
4
Very large shear stressAt square edge detail
“Normal” value(Scarf joint peak shear )
Analysis and DesignJoint Selection Considerations
P
P
P
P
Tension
23
• Scarf rate effects– “Lion’s share” of benefit to adherends occurs prior to increase to 20:1– Bond shear stress “peak to valley” amplitude increases with increasing length
• Yields surprising fatigue behavior (fatigue sensitivity manifests sooner in longer lap)
0
1
2
3
4
5
6
0 10 20 30 40 50 60 70 80 90Scarf Rate
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
Laminate = [90/45/0/-45/90/45/0/-45]s (Tape)
0.0
0.5
1.0
1.5
2.0
2.5
0 1Lap Position (x/L)
Loca
l Str
ess
/ Ave
rage
Str
ess
10:1 20:1 30:1
Analysis and DesignScarf Parametric Observations
P
P
P
P
Tension
24
• Scarf rate effects & damage tolerance considerations
0
1
2
3
4
5
6
0 10 20 30 40 50 60 70 80 90Scarf Rate
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
0
1
2
3
4
5
6
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00x/L
Baseline ~ Bond Shr / AVE Bond ShrBond Shr / lb. Applied LoadBond Shr / AVE Bond ShrAdherend Strain / Far Field Adherend Strain
Approx. 300% Increase Analytic Residual Str. Approx• 30:1 Scarf Joint• With Impact Damage• 50% Bond Loss (Center)• Implies 2/3 Strength Loss• Lap Length Consideration
For Damage Tolerance
Laminates[90/45/0/-45/90/45/0/-45]s (Tape)
Analysis and DesignScarf Parametric Observations
P
P
P
P
Tension
25
• Adhesive elastic modulus effects– Significant effect on adherend strain and peak-to-valley shear
stress in range of typical epoxy adhesives
0
1
2
3
4
0 50 100 150 200Adhesive Modulus (ksi)
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
Basis• 30:1 Scarf Rate• Constant Adhesive Thickness
0
1
2
3
4
0 50 100 150 200Adhesive Modulus (ksi)
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
Basis• 30:1 Scarf Rate• Constant Adhesive Thickness
Typical Epoxies
Laminates[90/45/0/-45/90/45/0/-45]s (Tape)
Analysis and DesignScarf Parametric Observations
P
P
P
P
Tension
26
• Adhesive thickness effects– Significant effect on adherend strain and peak-to-valley shear stress in range
of typical epoxy film adhesives (for single ply)
0
1
2
3
0.000 0.009 0.018 0.027 0.036 0.045 0.054Adhesive Thickness (in.)
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
Basis• 30:1 Scarf Rate• Constant Adhesive Modulus
Typical Epoxy Film Adhesives
0
1
2
3
0.000 0.009 0.018 0.027 0.036 0.045 0.054Adhesive Thickness (in.)
MAX Bond Shr / lb. Applied LoadMAX Bond Shr / AVE Bond ShrMAX Adherend Strain / Far Field Adherend Strain
Basis• 30:1 Scarf Rate• Constant Adhesive Modulus
Typical Epoxy Film Adhesives
Laminates[90/45/0/-45/90/45/0/-45]s (Tape)
Analysis and DesignScarf Parametric Observations
P
P
P
P
Tension
27
• Laminate stacking sequence (16 ply laminates)– Interchange 0º and 90º plies
• Increased max. bond stress, Decreased max adherend strainBond ~ [90/45/0/-45/90/45/0/-45]sBond ~ [0/45/90/-45/0/45/90/-45]s
Bond Shear Stress Profile
Substrate ~ [90/45/0/-45/90/45/0/-45]sSubstrate ~ [0/45/90/-45/0/45/90/-45]s
Adherend Axial Strain Profile
Analysis and DesignScarf Parametric Observations
P
P
P
P
Tension
28
Side Note: Current FAA WorkAnalysis / Damage Tolerance
• FAA collaborative research in process at NIAR• Scope: 1) analysis methods development, 2) bond
process robustness assessment, and 3) repair damage response
• Four major thrusts1. Baseline mechanical performance
– Static strength / post-fatigue residual strength / durability2. Degradation of mechanical performance as result of
contaminants on bond surface3. Impact damage site [in joint] resulting in greatest performance
degradation (i.e., What is critical site?)4. Detect-ability of bond surface contaminants
– Adequate pre-bond surface preparation / cleaning
Parent
Repair
Parent
Repair
Parent
Repair
Parent
Repair
29
Process Considerations
Surface Preparation
30
Surface PreparationScarf Grinding Operations
• Damages seldom occur in region with constant laminate thickness
• Ply drops in region effect final taper sand geometry• Sometimes difficult to anticipate what the final material
removal product should look like• Beneficial to have visualization of final taper sand prior to
starting grinding operations• During initial design, many airframe components modeled as
3-D entities in space and on A ply-by-ply basis• With appropriate software can generate 3-D simulation of
taper sanded region (prior to start)
31
Surface PreparationScarf Grinding Operations
• Three dimensional simulation enhances ability to produce repairs of higher precision
• Provides greater confidence that sanding operation was performed correctly
– Mechanic visual aid– Data for mechanizing major material removal steps
using robotic grinding• Hand finishing still required
– Improved records demonstrating that final repair geometry “looked like it was supposed to”
32
Surface PreparationPrecision Scarf Grinding
• Automated grinding using digital model data
Repair PliesDoubler Plies
Repair PliesDoubler Plies
33
Surface PreparationPre-Bond Surface Inspection
“Compact instrumentation” inspections methods– Contact angle (surface energy resulting from contaminants)– Near IR diffuse reflectance spectroscopy (shown)
Courtesy: Wichita State University, Department of Chemistry (FAA Collaboration)
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
40005000600070008000900010000
Abs
orba
nce
Wavenumbers (cm-1)
Synthesized “934” Resin
R2 = 0.9919
0
1
2
3
4
5
6
0 1 2 3 4 5 6
Actual
Cal
cula
ted
Water Uptake
( p )
R2 = 0.9882
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0.5 1.5 2.5 3.5 4.5 5.5
Actual
Cal
cula
ted
Water Desorption
34
Process Considerations
Laminate Patch Preparation
35
Process ConsiderationsLaminate Patch Preparation
• Even in repair scenarios with simple substrate geometry the patch plies seldom have regular, well defined edges
• A ply kit is created by….1. Taping clear Mylar over taper sanded region2. Hand tracing ply boundaries3. Transferring ply definition from Mylar4. Hand cutting repair plies with snips
• Final patch fit is marginal at best• The process is slow, and hand cutting
plies difficult to accurately accomplish• Randomness of patch fit adds to
randomness of structural performance
Example of Simple Taper Sand
36
Process ConsiderationsPrecision Laminate Patch Preparation
• Structural performance benefit in precision patch “ply kit”– Reduce randomness of structural performance
• Digital image processing is candidate technology to accomplish this
• With appropriate lighting, individual ply orientations can be detected using AFI instrumentation (Spirit AeroSystems patent pending)
• Points of prescribed density can be automatically defined in orientation change demarcation zones
• Splines laid through points to create ply edges• Feed data to automatic knife for precision cutting• Feed data to laser projector for patch ply lay-up• Provides permanent digital record of patch accuracy
37
Process ConsiderationsPrecision Laminate Patch Preparation
• Variation of lighting accentuates ply orientations
Spirit Aerosystems ~ Patent Pending
38
Process ConsiderationsPrecision Laminate Patch Preparation
• Examples: optical edge identification & digitization– Production of high precision “patch ply kits” possible
Spirit Aerosystems ~ Patent Pending
39
Process Considerations
Cure State Verification
40
Post Bond Cure State
• Resins are uniquely individual with regards to cure response to thermal cycle
– Some resins achieve an approximately full cure state early in thermal cycle
– In other resins cure reaction initiates late in cycle• Repairs frequently cured using heat blankets
– Heat blankets deliver thermal profile with tolerance band broader than autoclave cure
– Some areas hotter than others• Important to verify full cure
• DOT/FAA/AR-03/74, Office of Aviation Research, “Bonded Repair of Aircraft Composite Sandwich Structures”, John S. Tomblin, Lamia Salah, John M. Welch, Michael D. Borgman, 2004
• http://www.tc.faa.gov/its/worldpac/techrpt/ar03-74.pdf
41
Post Bond Cure State Inspection
• Hand held diffuse reflectance infrared spectroscopy– For assessment of cure state by peak amplitude area
11
Antares II
Remote Triggered Diffuse Reflectance Module
Integrating Sphere Reflectance Module
11
Antares II
Remote Triggered Diffuse Reflectance Module
Integrating Sphere Reflectance Module
Courtesy: Wichita State University, Department of Chemistry (FAA Collaboration)
42
Post Bond Cure State Inspection
• Cure state by diffuse reflectance IR spectroscopyCuring FM377U adhesive
0.00
0.20
0.40
0.60
0.80
1.00
400042004400460048005000
uncured41% cure98% cure
Wavenumbers (cm-1)
Abs
orba
nce
Curing FM377U adhesive
0.00
0.20
0.40
0.60
0.80
1.00
400042004400460048005000
uncured41% cure98% cure
Wavenumbers (cm-1)
Abs
orba
nce
0.00
0.20
0.40
0.60
0.80
1.00
400042004400460048005000
uncured41% cure98% cure
Wavenumbers (cm-1)
Abs
orba
nce
0.00
0.20
0.40
0.60
0.80
1.00
400042004400460048005000
uncured41% cure98% cure
Wavenumbers (cm-1)
0.00
0.20
0.40
0.60
0.80
1.00
400042004400460048005000
uncured41% cure98% cure
Wavenumbers (cm-1)
Abs
orba
nce
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)
Nor
mal
ized
pea
k ar
ea
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.2 0.4 0.6 0.8 1 1.2Extent of cure (α)
Nor
mal
ized
pea
k ar
ea
Courtesy: Wichita State University, Department of Chemistry (FAA Collaboration)
Chemometrics Calibration Curve for cure of FM377U adhesive
R2 = 0.987
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2Actual % cure
Cal
cula
ted
% c
ure
43
Post Bond Cure State Inspection
• Cure state by diffuse reflectance IR spectroscopyCuring T300/934 prepreg
0.00
0.20
0.40
0.60
0.80
1.00
1.20
400042004400460048005000
uncured42% cured94% cured
Wavenumbers (cm-1)
Abs
orba
nce
Curing T300/934 prepreg
0.00
0.20
0.40
0.60
0.80
1.00
1.20
400042004400460048005000
uncured42% cured94% cured
Wavenumbers (cm-1)
Abs
orba
nce
0.00
0.20
0.40
0.60
0.80
1.00
1.20
400042004400460048005000
uncured42% cured94% cured
Wavenumbers (cm-1)
0.00
0.20
0.40
0.60
0.80
1.00
1.20
400042004400460048005000
uncured42% cured94% cured
Wavenumbers (cm-1)
0.00
0.20
0.40
0.60
0.80
1.00
1.20
400042004400460048005000
uncured42% cured94% cured
Wavenumbers (cm-1)
Abs
orba
nce
Courtesy: Wichita State University, Department of Chemistry (FAA Collaboration)
Chemometrics Calibration Curve for cure of T300/934 prepreg
R2 = 0.983
35
45
55
65
75
85
95
105
35 45 55 65 75 85 95 105
actual % cure
Cal
cula
ted
% c
ure
44
Summary of Observations• For structurally critical bonded repairs to be pursued, absence of NDI
method for verifying bond strength creates need for high precision repairs using highly constrained processes performed by highly trained personnel (mechanics and analysts)
– Proof of statistical reliability only existing alternative for strength guarantee• Reliability must be demonstrated in terms of analysis & design,
surface preparation, patch “fit” precision, and process validation & records
• Bond-line failure mode may be consideration (or requirement) in lieu of traditional requirement for adherend failure mode
– Fairly repeatable / predictable mode (given adequate repair precision)– However, need A great deal more substantiation data to verify it as
reasonable alternative
45
Summary of Observations• Joint type strongly impacts peak bond stresses• Joint geometry parameters must be chosen with care
– Some parameters are “double edge sword”– For example scarf rate can improve static strength but, at the same time,
reduce fatigue resistance (increase sensitivity to fatigue)• Analysis methods must include laminate stacking sequence effects
– Including stiffness decay in taper region• Research on performance of damaged scarf joints is needed
– Propagation characteristics– Damage containment feature development and demonstration
• Candidate precision repair technologies exist but need maturation• Moisture detectable by diffuse reflectance near IR spectroscopy• Cure state detectable by diffuse reflectance near IR spectroscopy
46
Future• In any process control scenario reliability of success will
have to be proven• This reliability must be demonstrated through mechanical
testing the product from a LARGE volume of bondmentprocess cycles
• I have started this process at SPIRIT in support of one of our products, and have tested scarf joint strengths from approximately 200 bondment process cycles
– Many different mechanics– Many different cure thermal profiles– Many different resulting NDI scan quality results
• Have learned there are nuances to be considered in selecting a joint geometry for process reliability testing
– Joint shape and edge proximity (breathing during cure)
47
END